Why Your AI Assistant Still Feels Generic

About this Episode

Grok Bot, OpenClaw, Meta Muse, Instinct: every few weeks there's a new AI assistant that's supposed to run your life, and people keep jumping from one to the next. Brendan and Baird ask why they all still feel generic. Brendan's Claude Command Center has months of his data and still can't hand him the short to-do list a good assistant would. Baird's take is that the missing layer isn't more data. It's curation and intent: teaching the agent what actually matters. He walks through the agent he built in a day for work, wired into Granola, Gmail, Slack, Google Docs and Claude Code hooks, which turns everything he's doing into a simple board of what's on target and what's at risk. They compare notes on trying Instinct, call the consumer agents "the Squarespace of agents" (Jesse Hanley's word was "play school"), and land on the off-the-shelf vs. custom question Baird has faced his whole career, now that a Linear-style, MCP-enabled Kanban board takes one night to build. The back half covers why Opus 5.5 needs far less babysitting, why one-shotting an app is the same mistake as asking AI for a priority list from all your inboxes (either way you get a giant to-do list to sort through yourself), and why building in small increments keeps you in the decisions. As Baird puts it, that's collaboration, not just pressing approve.

Want help building an agent that's actually tailored to how you work? We're taking consulting work. Drop a comment or DM us.

Sponsor
Powerful podcast hosting & analytics
By podcasters, for podcasters. Since 2016
Brought to you by https://fireside.fm

claudecode #claudeai #aicoding #aiagents

Signup for our newsletter:
https://vibing-with-ai-code.beehiiv.com

Check out our Website
https://vibingwithaicode.com